High-Performance Multi-Stream Management for SSDs
Round 1
Reviewer 1 Report
Dear authors,
Interesting paper, few comments:
- several tables don't fit on the page, please fix that
- Verification by experiment, can that be added? Currently the verification is done digitally but any physically build SSD would truly provide the evidence your algorithm is working.
- I suggest to move the algorithm to an appendix
- The concept of lifetime as mentioned in the paper needs some further explanation
- Conclusions are not listed sufficiently, suggest to extend this part
The reviewer
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
This is a very nicely written and interesting paper.
I have few comments:
- (MI) Figure 1 is missing the x-axis title.
- (MI) In figure 1, the author should explain how you calculated the y – axis, i.e., Frequency of append writes (%).
- (MA) In line 114, the author states “In this paper, we propose a new multi-stream algorithm based only on the data types and operation types of the data to be stored.” While in line 129, the author states “Specifically, we propose classification of the user data based on their write operation types and not data types, namely synchronous-create, synchronous-append, and asynchronous.” These two statements appear to be contradictory to me. Please provide an explanation.
- (MI) Please specify how much is the error associated with your mean of intervals.
- (MA) In line 197, the author states “In our approach, after collecting a specific quantity of interval samples, we obtain a normal distribution based on the samples to predict the interval distribution of the subsequent append-writes.” Did the author confirm it with some tests to prove normality like KS Tests etc.?
- (OP) In line 302, the author states “In our experiments, the background GC is performed during idle time on the blocks with the highest number of invalid pages among blocks with at least 60% invalid pages”. Is there any specific reason to use this condition for background GC?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Dear authors,
You did a great job in improving the manuscript following the comments from the reviewers. For me, ready for publications.
The Reviewer